Date: Thu, 13 Aug 2009 18:38:48 +0200 From: Attilio Rao <attilio@freebsd.org> To: Damian Gerow <dgerow@afflictions.org> Cc: freebsd-current@freebsd.org, Robert Noland <rnoland@freebsd.org> Subject: Re: No display after resume in r196086 Message-ID: <3bbf2fe10908130938v525a18c3p9c4e10db662ab3c0@mail.gmail.com> In-Reply-To: <20090813162705.GA1468@plebeian.afflictions.org> References: <20090813002104.GA1481@plebeian.afflictions.org> <1250161757.1823.18.camel@balrog.2hip.net> <3bbf2fe10908130652t77767929q44ba12f39cb3f997@mail.gmail.com> <20090813162705.GA1468@plebeian.afflictions.org>
next in thread | previous in thread | raw e-mail | index | archive | help
2009/8/13 Damian Gerow <dgerow@afflictions.org>: > Attilio Rao wrote: > : why don't you directly try the revision immediately before newbus > : locking (so immediately prior to r196037) and try to verify it still > : works. > > Culprit confirmed: I can resume with r196036, but not r196037. > > : I'm aware of at least one possible deadlock source but it should not > : happen while suspend/resume. However, I'm preparing an errata patch > : for newbus locking. > : If you verify newbus locking is the real culprit and that my further > : patch doesn't fix your problem we can try to hammer it down further. > > Let me know what else you might need. So, what you can do is to compile a kernel with the following options: KDB, DDB, INVARIANT_SUPPORT, INVARIANTS, WITNESS when you get the deadlock, you should break in DDB. Through a textdump(4) (or serial console or whatever you prefer) you should get the following informations: bt, ps, show allpcpu, alltrace, show alllocks and report here. Thanks, Attilio -- Peace can only be achieved by understanding - A. Einstein
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3bbf2fe10908130938v525a18c3p9c4e10db662ab3c0>